A Thousand Games In, It Still Doesn’t Even Out
Is a program’s all-time record skill, or could luck have done it? Here is the answer, with its number. Seventy-four programs have played 1,000 or more major games. Luck could explain the records of 20. For the other 54, it can’t.
That should surprise you, because you have heard the opposite your whole life: it all evens out. Over enough games, people figure, the bad bounces cancel the good ones and everybody drifts toward average. For a coin, that is true. For a football program, it mostly isn’t.
By the end you will be able to look at any record and ask the one question that tells you how much luck is in it. How many games?
Watch Every Program at Once
Start with the picture, because it does the arguing for us. Open Every Program Since 1869 and press play. Each dot is one program. It moves right as the program plays more games, and up or down with its winning percentage. In 1869 there are two dots. By 2025 there are 304, and 168 of them have stopped: schools that dropped down a level, merged, or quit, plus the wartime service teams of the 1940s.
The idea is borrowed from Julian Critchlow, whose map of English Channel swims starts every swimmer at hour zero. Our page has the same switch. Pick “From season 1” and every program kicks off together, Yale in 1872 beside Boise State in 1996.
What a Coin Would Do
Picture a team that wins every game with a coin flip. After 10 games, its record could easily be anywhere from .200 to .800. Ten flips can do almost anything. After 1,000 games, it would almost never leave .470 to .530. Draw those ranges for every number of games and you get the tan funnel in the chart above: wide where there are few games, narrow where there are many.
Notice what that means for the left side of the chart. Randolph Field, a 1944 service team, played 12 games and lost none. A perfect record! But 12 games is the wide part of the funnel. The record cannot tell you how much was skill, so a perfect record there tells you less than it seems.
Now look right. Alabama is .750 after 1,360 games. A coin would need more luck than the world has ever seen to do that. The same goes for Ohio State at .758, Michigan at .732 and Notre Dame at .731. Down at the bottom, Wake Forest sits at .419 after 1,219 games. That isn’t bad luck either. It’s a long, real record.
Ten games, and luck can make almost any record. A thousand, and luck can barely move it.
Who Could Pass for a Coin
Twenty programs with 1,000 or more games sit inside the funnel. Virginia is the closest: .499 after 1,276 games. Illinois is .504 after 1,321. Purdue, Baylor, Kentucky, Duke and South Carolina are all within a whisker of .500. That does not mean those programs were coins; good decades and bad decades can cancel out. It means their records alone can’t tell you apart from one.
And here is the twist. The funnel catches a fair coin 19 times in 20, not every time. So even if all 74 programs were coins, three or four would land outside it by luck. Fifty-four do. So the spread you see is real. Some places have simply been better at football for a century.
Then There Is Indiana
Indiana is .426 after 1,239 games, the third-lowest of the 74. By the funnel, that long losing record is not bad luck. And last season Indiana went 16–0 and beat Miami 27–21 for its first national title. (We looked at how its turnaround began in Issue #207.)
Both facts are true at the same time. The all-time record measures a program across 127 seasons, many coaches and thousands of players. It describes the past. It says very little about this year’s team, which is one season: the wide part of the funnel all over again.
What to Take Home
Before you believe a record, ask how many games it rests on. With a few games, luck can make almost anything, so hold the record loosely. With a thousand, luck can barely move it, so whatever you see is real.
And retire “it all evens out.” That is a rule for coins. Teams are not coins, which is the whole reason anybody watches.
How this was measured, and what it cannot do
Records: James Howell’s College Football Historical Scores, 304 programs and 14,408 seasons from 1869 to 2025. These are results as played on the field: vacated wins stay in, so Alabama is 999–319–42 here against an official 985–345–43. Howell counts only the seasons a program played at the major-college level, so Notre Dame enters in 1899 and Yale leaves after 1981. Totals will not match a school’s own record book.
Checked against CollegeFootballData for 2005–2025: 31,525 of 31,553 regular-season games match to the exact score, and none of the rest changes who won. Yale’s 1872–1981 record differs from Sports-Reference’s by one game.
The funnel is .500 plus or minus two standard deviations of a fair coin, one over the square root of the games played. A tie counts as half a win. Leaving ties out moves no program across the line at 1,000 games.
What it cannot tell you: a coin is the simplest kind of luck. Real luck in football is lumpier (injuries, schedules, a bad decade), so the funnel is a floor, not a full account. And it says nothing about why a program was good, only that it was.
Argued about something you could not settle? Send it to us · what makes a good question